Studies in several cell types indicate that the actions of integrin receptors for extracellular matrix and receptors for growth factors are synergistic in regulating cellular differentiation and function. We studied the roles of the α1β1 and α2β1 integrin collagen receptors in regulating the differentiation of 2T3 osteoblastic cells in response to bone morphogenetic protein (BMP)-2. The immortalized 2T3 cell line was established from the calvaria of mice transgenic for a BMP-2 promoter driving SV40 T-antigen. These cells require exogenous BMP-2, as well as ascorbic acid and β-glycerolphosphate, for expression of a mature osteoblast phenotype and formation of a mineralized matrix. To determine how integrin receptors for collagen-I affect BMP-2 signaling, function-perturbing anti-rat α1 and/or α2 integrin subunit, or anti-type I collagen (Col-I), antibodies were added to human recombinant (hr)BMP-2–treated 2T3 cultures at confluence (C0) or at 4 or 8 days postconfluence (C4, C8). After 4 days of exposure to the antibodies, cultures were assayed for alkaline phosphatase (ALP) mRNA levels and enzyme activity and for cAMP production in response to parathyroid hormone. Addition of anti-Collagen-I or both anti–integrin-α1 and -α2 antibodies to C0 cultures blocked expression of these early osteoblast markers by more than 90%, and also blocked mineralization (0.5–1.8% control) of these cells. In all cases, adding anti-α1 or anti-α2 antibodies separately produced partial effects, while their combined effect approached that of anti-Collagen-I. When antibodies were added to more differentiated 2T3 cells, the inhibitory effects decreased. 2T3 cells carrying constitutively active BMP receptor (caBMPR-IB) showed elevated ALP activity without hrBMP-2; this constitutive activity was also suppressed by α1 and α2 integrin antibodies and by anti-Col-I antibody. Together, our data suggest that a signal(s) from collagen integrin receptors regulates the response to BMP downstream of BMPR-IB and upstream of the regulation of ALP mRNA and other early markers of osteoblast differentiation.
During early stages of differentiation, osteoblasts initiate synthesis of an extracellular matrix (ECM) consisting primarily of type I collagen (Col-I). Then they produce alkaline phosphatase (ALP) and a variety of characteristic noncollagenous proteins, followed by induction of ECM calcification.(1–3) The ECM plays important roles not only as a field for calcification in bone tissue but also as a regulator of osteoblast differentiation. Several studies have demonstrated that osteosarcoma cells, calvarial osteoblasts, and preosteoblasts show increased expression of ALP and other markers of the osteoblastic phenotype when plated on Col-I as compared with tissue culture plastic.(4–10) Other studies have shown that expression of ALP depends on the production of Col-I.(11,12)
Integrins are the likely transducers of many of the signals from ECM that regulate osteoblast commitment and/or differentiation. Integrins are composed of noncovalent αβ heterodimers. Among the more than 20 different integrins,(13) it is known that the α1β1 and α2β1 complexes are collagen receptors.(14–20) Studies using function-perturbing integrin antibodies have shown that β1 or α2β1 integrin signaling is required for induction of ALP in human osteosarcoma cells or mouse MC3T3-E1 cells.(21,22) Thus, ECM and growth factors are likely to act cooperatively to stimulate osteoblast differentiation, as has been shown for other cell types. For example, fibroblast growth factor and activin A are potent mitogens of P19 mouse embryonal carcinoma cells when the cells are grown on laminin- and fibronectin (FN)-coated dishes but not on plastic.(23) Proliferative responses of murine mammary carcinoma cells to platelet-derived growth factor and basic fibroblast growth factor are dependent on adhesion to FN.(24) Fibroblasts in a collagen gel proliferate less in response to platelet-derived growth factor than they do when plated on tissue culture plastic.(25) In preosteoblast cells, Col-I increases calcitonin and parathyroid hormone (PTH) receptor-mediated signal transduction.(26)
Bone morphogenetic proteins (BMPs) were originally purified on the basis of their ability to induce ectopic bone formation.(27,28) Many studies have shown that BMP-2 stimulates the osteoblast differentiation of bone-marrow-resident pluripotent cells, C2C12 osteoblast progenitor cells, and cultured rat calvarial osteoblasts.(5, 29–34) Moreover, the fact that BMP mRNAs localize at sites of bone formation or fracture repair suggests that BMPs have essential roles in differentiation and maintenance of bone cells.(35–38) A question that arises from these observations is whether cell–ECM interactions through integrins are required for osteoblast differentiation induced by BMP.
Recently, the 2T3 osteoblast clonal cell line was established from calvaria of transgenic mice carrying a BMP-2 promoter–SV40 T-antigen construct.(39) This is a tissue-specific weak promoter that allows derivation of immortalized but nontumorigenic cell lines capable of extensive differentiation. Addition of human recombinant (hr)BMP-2 strongly enhances ALP activity, cAMP production in response to PTH, and ECM calcification in these 2T3 cultures. Thus, 2T3 cells go through an ordered program of early osteoblast differentiation followed by production of an extensive mineralized matrix, and are therefore suitable as an osteoblast differentiation model for clarifying the roles of integrins in the response of 2T3 cells to BMP-2 action. We now show that signals from collagen integrin receptors are required for osteoblast differentiation induced by BMP-2. Function-perturbing antibodies for α1 and α2 integrins and antibody against Col-I blocked initiation of BMP-dependent osteoblast differentiation. The inhibitory effects of the antibodies decreased when they were added to more differentiated 2T3 cells. Moreover, 2T3 cells carrying constitutively active BMP receptor (BMPR)-IB showed up-regulated ALP activity without hrBMP-2. Function-perturbing α1 and α2 integrin antibodies or anti-Col-I antibody blocked the signals from constitutively active BMPR-IB, indicating that signals transduced by these integrins affect 2T3 response to BMP-2 receptor activity.
MATERIALS AND METHODS
2T3 cells were seeded at a density of 1500 cells/6-mm well in alpha minimum essential medium (α-MEM) supplemented with 7% fetal calf serum (FCS). After cultures reached confluence on day 4, cells were maintained in α-MEM supplemented with 7% FCS, 100 μg/ml of ascorbic acid, and 5 mM β-glycerophosphate, with or without hrBMP-2 (Genetics Institute, Boston, MA, U.S.A.). Day 4, day 8, day 12, day 16, and day 20 of the culture are hereafter designated as postconfluence day C0, C4, C8, C12, and C16, respectively.
Constitutively active BMPR-IB–transfected 2T3 cells
2T3 cells were transfected with a mutated constitutively active human type IB BMP-2 receptor subunit, and stable transfectants were selected as described by Chen et al.(40) and cultured as above.
Hamster anti-rat integrin α1 and α2 function-perturbing monoclonal antibodies (MAbs) (Ha 31/8 and 1/29) were described previously.(41,42) These antibodies in combination inhibited undifferentiated 2T3 cell adhesion to collagen in serum-free medium containing 0.2% bovine serum albumin (data not shown). Rabbit anti-mouse FN antibody, rat anti-mouse α5 antibody and goat anti-Col-I antibody were purchased from Life Technologies (Grand Island, NY, U.S.A.), Pharmingen (San Diego, CA, U.S.A.), and Southern Biotechnology Associates, Inc. (Birmingham, AL, U.S.A.), respectively. In the case of the MAbs, the immunoglobulin G (IgG) fraction was enriched from conditioned medium of the hybridoma by ammonium sulfate precipitation. The IgG fraction was purified from polyclonal antisera using Protein A. Function-perturbing activity of anti-collagen and anti-α1 and anti-α2 antibodies was verified using 2T3 cells in cell attachment assays using established procedures.(43)
ALP activity was measured as described previously, using p-nitrophenyl phosphate (pNP) as a substrate.(44) Briefly, cell layers were homogenized in 0.9% NaCl/0.2% Triton X-100. Aliquots were incubated with 0.5 M Tris-HCl (pH 9.0) containing 0.5 mM pNP and 1 mM MgCl2 at 37°C for 30 minutes. Absorbance was measured at 405 nm. One unit was defined as the activity catalyzing the hydrolysis of 1 μmol of pNP/mg of protein/30 minutes.
Alizarin red staining and calcium content
On C16, the cell layer was washed with phosphate-buffered saline and water, quickly followed by 100% ethanol for fixation. For alizarin red staining, cell layers were stained with 1% alizarin red S solution for 30 minutes and washed with distilled water several times. For measurement of calcium content, calcium in cell layers was extracted with 50 μl of 0.1 N HCl for 20 minutes, and then calcium content was determined by a Sigma diagnostic kit for calcium (Sigma, St. Louis, MO, U.S.A.).
On C4 or C8, medium was replaced with α-MEM containing 0.7% FCS and 3 × 10−8 M of bovine 1–34 PTH (Sigma). Cells were incubated for 5 minutes, after which cold 5% trichloroacetate was added to extract intracellular cAMP. Aliquots were washed five times with water-saturated ether, and then cAMP content was measured with a cAMP enzyme immunoassay system (Amersham Life Science, Inc., Arlington Heights, IL, U.S.A.).
Total RNA from C4 cultures was purified with Qiagen RNeasy total RNA system (Qiagen, Inc., Santa Clarita, CA, U.S.A.). Fifteen micrograms of total RNA was applied to 1% agarose gels and transferred to a Duralon-UV membrane (Stratagene, La Jolla, CA, U.S.A.). Two different filters were hybridized with the ALP probe. cDNA for rat ALP was kindly provided by Dr. G.A. Rodan, Merck Research Laboratories. Radiolabeled probes were made using a Random Primed DNA labeling kit (Boehringer Mannheim, Indianapolis, IN, U.S.A.).
Immunoprecipitation of integrins
2T3 cells were cultured with or without 30 ng/ml of hrBMP-2 for 48 h postconfluence (C0–C2). Then cells were surface labeled with biotin and lysed as described previously.(44,45) Two hundred micrograms of aliquots of protein were immunoprecipitated with excess anti-α1 and anti-α2 antibodies (Ha 31/8, Ha 1/29), followed by anti-hamster IgG bound to agarose beads. The goal was to precipitate all antigen with a single round of primary antibody and secondary antibody(45) so as to compare integrin levels in the presence and absence of BMP-2. After washing, immune complexes were treated with SDS sample buffer and applied to 7% SDS-polyacrylamide gels, separated, blotted, and visualized by enhanced chemiluminescence according to the manufacturer's instructions (Amersham).
hrBMP-2 stimulates osteoblastic differentiation of 2T3 cells
2T3 cells reached confluence on day 4 (C0). Cells were then treated with various concentrations of hrBMP-2 for the next 4 days, and ALP activity was determined at C4. As shown in Fig. 1A, hrBMP-2 induced ALP activity in 2T3 cells in a dose-dependent manner. When cells were cultured continuously with 30 ng/ml of hrBMP-2 until C16, ALP activity increased, peaking on C8 at a level comparable to that of spontaneously differentiating rat primary calvarial osteoblasts (Fig. 1B and data not shown). Calcium content of the cultures increased from C8 onward, reaching 579 ± 40 μg/mg of protein (Fig. 1C). Extensive calcium deposition could be seen in the hrBMP-2–treated cultures, as demonstrated by alizarin red staining (Fig. 1). However, neither ALP activity nor matrix calcification of 2T3 cells could be observed in the cultures without hrBMP-2 (Figs. 1A–1D). These data demonstrated that hrBMP-2 was essential for 2T3 cells to manifest an osteoblastic phenotype, as described previously.(39)
α1β1 and α2β1 integrins are necessary for BMP-2 induction of both early and late osteoblast functions
To determine the involvement of α1 and α2 integrin during 2T3 osteoblast differentiation induced by hrBMP-2, we added function-perturbing anti-integrin antibodies to cultures for 4 days, starting at C0, C4, or C8, and assayed for ALP activity or cAMP production in response to PTH according to the schedule shown in Fig. 2. Addition of 100 μg/ml of anti-α1 and/or anti-α2 integrin antibodies or 3 μg/ml anti-Col-I antibody starting at C0 or C4 blocked the BMP-2 stimulation of ALP activity (Fig. 3A) and cAMP production in response to PTH (Fig. 4A). The decrease in ALP activity was associated with a reduced ALP mRNA level, as assessed by Northern blot analysis on cells at C4 (Fig. 3B). Anti-α1 and anti-α2 integrin antibodies had partial effects on ALP activity when added individually (Figs. 3 and 4). When added together, however, their effects approached that of anti-Col-I antibody. The inhibitory effect of all of the antibodies declined with osteoblast differentiation (Figs. 3C and 3D and Fig. 4B). In contrast to results with anti-Col-I and anti-integrin collagen receptors, interfering with 2T3 cell interactions with FN by treating cultures with anti-α5 integrin antibody and anti-FN antibody had weak effects on differentiation at all times tested; addition of these antibodies to cells at C0, C4, or C8 had little effect on the induction of ALP or the response to PTH (Fig. 5). None of the antibodies used in these experiments affected cell numbers in cultures significantly (data not shown).
Next, we determined the roles of α1 and α2 integrins in ECM calcification. Cells were treated with various antibodies from C0, C4, or C8 until C12 or C16 according to the schedule described in Fig. 6. Figure 7 presents results of alizarin red staining and calcium content analysis at C16. Anti-α1 and/or anti-α2 antibodies or anti-Col-I antibody blocked calcification and formation of a mineralized matrix by more than 95% when added beginning at C0, with decreasing effects when added beginning at C8 (70–100% control) (Fig. 7 and data not shown). Anti-α5 and anti-FN antibody had no effect on calcification when added at C0 or C8 (data not shown). These data clearly demonstrate that collagen integrin receptors are essential at early stages of the response of 2T3 osteoblastic cells to BMP-2.
BMP-2 increases surface expression of α1β1, but not α2β1
To understand how collagen integrin signals might regulate early stages of osteoblast differentiation initiated by BMP-2, we added 30 ng/ml of hrBMP-2 to the culture beginning at C0 for 2 days and measured changes in α1β1 and α2β1 integrin expression in 2T3 cells by quantitative immunoprecipitation. As shown in Fig. 8, the level of α1β1 integrin protein was consistently enhanced by about 2-fold following hrBMP-2 treatment; however, BMP-2 had only a marginal effect on α2β1 protein expression in 2T3 cells.
Integrin collagen receptors block BMP-2 response downstream of BMP receptors
To pursue the possibility that signals from integrin collagen receptors interfere with BMP-2 receptor activation, we used 2T3 cells stably transfected with constitutively active BMPR-IB.(40) This stable cell line spontaneously differentiates and forms a mineralized matrix in the absence of BMP-2. Addition of function-perturbing α1 and α2 integrin antibodies or anti-Col-I antibody at confluence blocked the increase in ALP activity on C4 (Fig. 9), indicating that the signals from integrin collagen receptors affect the BMP-2 response downstream of BMP-2 receptor activation.
Evidence is accumulating that integrins can collaborate with growth factor receptors and modulate their signaling.(46–49) In osteoblasts, signals from Col-I are reported to promote differentiation in response to transforming growth factor (TGF)-β/BMP family members.(4,6–10) Because hrBMP-2 initiates and supports terminal differentiation of 2T3 cells, using this cell line allowed us to investigate the roles of collagen receptor integrins on BMP-2–dependent osteoblast differentiation. Our study clearly showed that perturbing the functions of both α1β1 and α2β1 integrins, or of Col-I, blocked 2T3 osteoblast differentiation induced by hrBMP-2. BMP-2–induced elevation of ALP mRNA and enzyme activity, as well as enhanced cAMP production in response to PTH, was strongly suppressed. The effects of these antibodies on late events in 2T3 cell differentiation (e.g., calcification of ECM) were weak when they were introduced after day C4. Therefore, signals from Col-I transduced via α1β1 and α2β1 integrins likely function at early stages of the 2T3 cell differentiation response to BMP-2. It is unlikely that antibodies against α2 and α1 integrins or Col-I lacked access to sites of ECM–integrin interaction at later stages, since we found, using confocal microscopy, that the anticollagen and anti-integrin antibodies penetrated throughout the entire cell layer in these cultures (data not shown).
In the present study, we used Ha 31/8 and Ha 1/29 for functional perturbation of α1 and α2 integrins in murine 2T3 cells. Although these MAbs were raised against rat integrins, Yao et al.(50) and Helfrich et al.(51) reported that these antibodies recognize and functionally perturb mouse α1 and α2 integrins as well as rat integrins. Blocking either α1 or α2 alone had partial effects, indicating that both of these integrins participate in the regulation of the response of 2T3 cells to BMP-2. The two integrin α subunits are structurally similar and are redundant or cooperative in some of their activities.(52,53) For example, Lee et al.(53) showed that α1 and α2 integrins cooperatively mediate adhesion to collagen in vascular smooth muscle cells. These observations and our results support the idea that α1 and α2 can function cooperatively not only as adhesion receptors but also as signal-transducing receptors in osteoblasts.
Several mechanisms could explain how integrin–Col-I interactions regulate the 2T3 cell response to BMP-2. One possibility is that anchorage via integrins to collagen-I is required for successful BMP-2 signal transduction. This hypothesis predicts either that the BMP-2 signal transduction pathway is interrupted when integrin–Col-I interactions are perturbed, or that one or more BMP-2–responsive genes cannot be expressed in the absence of cooperative signals transduced by integrin collagen receptors. An alternative explanation is that the BMP-2 signaling pathway is transduced normally, but integrin–Col-I interactions are required for the optimal function of gene(s) expressed in response to BMP-2.
In initial experiments to distinguish among these possibilities, we found that blocking integrin–Col-I interactions suppressed the increase in ALP mRNA triggered by BMP-2. This suggests that anchorage via integrin collagen receptors regulates events at, or upstream of, the synthesis or stabilization of mRNA for BMP-2–responsive genes. Whether the effect of BMP-2 signaling on expression of ALP is direct or indirect is not yet clear. The latter mechanism is likely, however, as BMP-2 regulates expression of the recently cloned osteoblast-specific trans-activating factor Cbfa-1/Osf-2(54) in 2T3 cells.(40) This factor, in turn, activates the group of osteoblast marker genes, including ALP, whose regulatory regions contain the OSE-2 cis-element recognized by this factor.(54)
Next, we examined effects of perturbing integrin–Col-I interactions on the initial steps of the BMP-2 signal transduction cascade. Members of the BMP family exert their actions through two types of serine/threonine kinase receptors by binding directly to and activating the type II receptor, which then activates the type I receptor.(55,56) Thus, BMP-2 binds BMPR-II, which can activate BMPR-IA or BMPR-IB.(57–60) 2T3 cells expressing constitutively activated (ca) forms of BMPR-IA or BMPR-IB have been established. Those expressing caBMPR-IA differentiate spontaneously along the adipocyte pathway, while 2T3 expressing caBMPR-IB exhibit spontaneous osteoblastic differentiation.(61) Therefore, we asked whether antibody perturbation of integrin–Col-I interactions could block the constitutive expression of elevated ALP activity and responsiveness to PTH present in 2T3 cells expressing caBMP2R-IB. This constitutive response was blocked, indicating that the integrin-dependent step(s) in the 2T3 cell response to BMP-2 are downstream of the sequential activation of type II and type I receptors by BMP-2. Effects on the next steps in this signal cascade, which include activation of members of the Smad family and translocation of Smad complexes to the nucleus,(62) are under investigation.
An alternative mechanism for integrin-mediated effects on BMP-2–induced osteoblast differentiation arises from the requirement for ascorbic acid, as well as BMP-2, for optimal differentiation of osteoblasts, including 2T3 cells. Collagen integrin receptors appeared to be required in osteoblast differentiation induced by ascorbic acid.(22) Moreover, studies using antisense cDNA showed that focal adhesion kinase is involved in this process.(63) Recent studies(64) suggest that collagen matrix synthesis in response to ascorbate is a necessary component of the osteoblast differentiation program that is regulated independently of the response to BMP-2.
Since TGF-β/BMP superfamily members have been shown to stimulate expression of integrins and ECM components in mesenchymal cells and osteoblasts,(65–68) we also determined effects of BMP-2 treatment on expression of integrin collagen receptors. BMP-2 treatment resulted in a modest increase in the level of α1β1, but not α2β1, complexes in 2T3 cells. However, since 2T3 cells express both receptors before and after BMP-2 treatment, the enhanced α1 integrin expression is not likely to be a critical part of the mechanism by which integrin collagen receptors regulate effects of BMP-2 on osteoblastic differentiation of these cells. Whether BMP-2 affects integrin activation is under investigation.
Interestingly, interfering with 2T3 cell interactions with FN by using either anti-integrin α5 antibody or anti-FN antibody did not block BMP-2 induction of 2T3 cell differentiation, suggesting some specificity in the effects of integrin–ECM interactions on BMP-2 signaling in this system. In contrast, addition of anti-FN antibody or FN fragments from the central cell-binding region, or anti-integrin FN receptor antibodies, inhibited formation of bony nodules and suppressed expression of ALP and osteocalcin mRNA in cultures of fetal rat primary calvarial osteoblasts.(69,70) These data indicate that FN provides crucial signals for differentiation of primary osteoblasts. Since the primary osteoblast system is not dependent on exogenous BMP-2 for differentiation, it is difficult to compare results in the two systems. However, one hypothesis that could reconcile the apparently contradictory results regarding the importance of interactions with FN for osteogenesis is that anchorage to FN and anchorage to Col-I are both important, but function at different stages of the osteoblast differentiation program. In this scenario, osteoblast interactions with FN are critical at a point upstream from BMP-2 signaling and osteoblast–collagen interactions. Therefore, treatment of 2T3 cells with BMP-2 would bypass the requirement for FN.
In summary, we conclude that collagen integrin receptors function at early stages of BMP-dependent osteoblast differentiation in 2T3 cells. In this BMP-2–dependent system, signals from collagen through collagen integrin receptors appeared to be more important than signals from FN. Our data thus far indicate that integrin collagen receptors are critical at step(s) downstream of BMPR-I activation and upstream of the transcriptional or post-transcriptional regulation of expression of mRNA for ALP, and perhaps other early markers of the osteoblastic phenotype.
We are grateful to the Genetics Institute for hrBMP-2, and to Dr. G. Rodan for rat ALP cDNA. We thank Evangeline Leash for expert editing of the manuscript. This work was funded by National Institute of Dental and Craniofacial Research/National Institutes of Health grant no. P50DE10306.